Research scientists have known for decades that biological organisms use carbohydrates in a number of ways – including as an energy source and as structural material. They’ve also recognized that proteins and cell surfaces are loaded with sugar groups – although they once thought that these moieties were merely decoration and served no important function. Well, they couldn’t have been farther off base. As it turns out, carbohydrates are vital to life itself: Without them, cells couldn’t differentiate, develop, grow, move or communicate. Understanding exactly how they do this is the realm of glycobiology – and important new technologies being developed by a cadre of young firms are going to be critical components of the quest.
Raymond Dwek didn’t start out studying carbohydrates – yet today he’s widely regarded as the father of glycobiology. Back in the 1980s, however, Professor Dwek and his colleagues at the University of Oxford were working on antibodies – when one day they discovered that the antibodies produced by rheumatoid arthritis patients differed from those produced by unaffected individuals in one important respect: The carbohydrate groups attached to the antibody molecules weren’t the same. As it turns out, the culprit was a faulty enzyme, whose normal function is to attach the sugar galactose to the appropriate sites on proteins. Many researchers might have cringed at this finding, knowing full well that sugars were impossibly difficult to study, and left it at that. But this one observation was enough for Dwek. “It gave us a marker,” he explained. “Through that we developed the technology” that has literally revolutionized the study of glycobiology -- a term that Dwek coined in 1988.
That same year he founded Oxford GlycoSciences plc to identify and analyze glycoproteins. The company, which was recently acquired by Celltech Group plc, ended up doing a lot more than that, though: It also developed a drug for treating the rare genetic lipid storage disorder Gaucher disease. The product, Zavesca, is a small molecule inhibitor of glucosylceramide synthase, the enzyme responsible for the first step in the synthesis of most glycolipids; it was approved in the EU in November 2002, in Israel in June 2003, and in the U.S. in July 2003.
Dwek also founded the University of Oxford’s Glycobiology Institute back in 1988 – and today “we have a big institute and a huge number of people want to come [work here]. There are over 100 people on the waiting list,” he said.
That’s no surprise, for the Institute is a veritable hotbed for carbohydrate-based drug discovery. For instance, Dwek’s group created a class of small molecule glycomimetics called iminosugars that proved capable of combating infection in vitro by flavoviruses, including hepatitis B and hepatitis C. The iminosugar blocks the virus’ iron channel, which interferes with its assembly and makes it non-infectious, Dwek explained. United Therapeutics Corp. obtained the exclusive worldwide rights to these compounds, and is currently conducting Phase II trials in patients with HCV infection.
Dwek added that there are other uses for iminosugars, as well – including a possible role in contraception. In animal studies, at least, researchers have found that iminosugars “act on a protein in sperm, making it dysmorphic. That inhibits the binding of the sperm to the egg. This could be a reversible sterility pill for males.”
But iminosugars are merely the latest in a long list of discoveries that have emanated from the Glycobiology Institute. Its evolution – like that of the entire field of glycobiology – has been remarkable.
Two decades ago, the chemistry and function of sugars were one of biology’s famous black boxes, and few researchers were willing to tackle carbohydrate biochemistry – especially because the technology to analyze these extremely large and complex molecules simply wasn’t available.
Moreover, although it was generally recognized that biological organisms use carbohydrates in various ways (e.g., as an energy source and structural material), dogma held that the sugar groups on proteins and cell surfaces were no more than decoration – and certainly served no important function.
Well, that interpretation turned out to be distinctly not true. Au contraire, researchers now know that these nearly ubiquitous carbohydrates – which constitute more than 90 percent of all cell surface antigens and are present on more than 80 percent of all known proteins -- play a critical role in many cellular functions. They’re involved in cellular communication, growth, motility, differentiation and development. They regulate the activity of immune cells and proteins. They recognize invading microbes and help the immune system distinguish self from non-self. Glycosylation is important to receptor binding and signaling pathways. Malfunctions in carbohydrate synthetic pathways can lead to devastating diseases (such as Gaucher disease, mentioned earlier). And differences in the cell surface patterns of carbohydrates distinguish cancerous cells from normal ones.
As well, techniques for analyzing complex carbohydrates – long chains or branched structures containing up to 200 or more monosaccharide units, linked together in a number of possible ways – are finally coming into their own. Aided by advances in mass spectrometry (MS) and nuclear magnetic resonance (NMR), studies on structure-activity relationships and sophisticated bioinformatics, researchers have launched an assault on these heretofore mysterious molecules.
It’s not like studying DNA or proteins, though: Although it’s now possible to sequence carbohydrates on gels, there are no automated sequencers for large-scale efforts and no amplification enzymes to easily generate sufficient quantities for experimental purposes. Synthesis is still tough, too, though the first automated synthesizer was developed at MIT last year.
As if there weren’t enough challenges, it’s also now clear that glycosylated proteins are not homogeneous. Any one purified protein actually exists in a number of versions, called glycoforms, which exhibit differences in the exact sugars that are attached at the various binding sites. That means that a protein cannot be defined solely on the basis of its amino acid sequence, because it’s not actually a single species. But knowing this allows researchers to investigate the activities of various glycoforms, which should lead to a better understanding of a therapeutic protein’s efficacy – and consequently, to designing better, more potent drugs.
Armed with a clearer understanding of sugars’ roles, plus improving technologies to analyze them, glycobiology is entering a new era. According to Oxford’s Dwek, “Glycobiology is at the crossroads.” But going forward, “it won’t be just any old glycobiology.” Conventional approaches won’t work. “Scientists have still got to be inventive and understand what they’re doing. A lot of it will be luck.”
“Inventive” certainly describes Momenta Pharmaceuticals Inc., one of the new crop of glycobiology companies. Founded in 2001 on technology developed at MIT, the privately held company is perfecting a high throughput method for the precise chemical characterization of complex sugars – both linear and branched. The Cambridge, MA firm’s technology involves breaking up these huge molecules into smaller fragments with specific enzymes, sorting the bits with high-pressure liquid chromatography and then characterizing the individual units within each piece by mass spectroscopy and NMR. With the help of sophisticated computer algorithms that can simulate the massive number of theoretical sequences that might belong to the complex sugar being investigated, and an iterative experimental process that systematically eliminates the possibilities, Momenta’s technology is capable of deriving the full sequence of the original complex.
According to chairman and CEO Alan Crane, “Momenta’s technology has three components: proprietary enzymes, equivalent to restriction enzymes, that chop sugars in very specific places; improvements on existing analytical technology (both MS and NMR), which have been adapted for use with sugars; and bioinformatics.” Moreover, the technology requires very small amounts of starting material and can even be used to analyze a mixture of different polysaccharides.
The company’s enzymes, all licensed from MIT, are absolutely critical for determining structure. The collection contains naturally occurring enzymes (such as heparin-degrading enzymes isolated from bacteria) as well as about 120 mutant enzymes with altered specificities that degrade polysaccharides at different places, explained Ganesh Venkataraman, Momenta’s VP of technology. The firm also has various enzymes that attach or remove side groups on the sugar molecules as well as transferases that it can use to build sugars, he added.
“Our technology allows us to sequence a sugar of any length, which used to be impossible.” For instance, using older methods, the time and labor needed to sequence even a hexasaccharide would be enough to constitute a graduate thesis, Venkataraman said. “Now, it takes 15 minutes.”
Unidentified heparin chain. Image courtesy Momenta Pharmaceuticals Inc.
That’s good, because complex sugars can be much larger. Momenta has sequenced both deca- and dodeca-saccharides. And here, the possible combinations are staggering: 255 million for the 10-unit sugar and more than 12 billion for the 12-unit sugar. Why is this so? Because not only are there a number of different ways for monosaccharides – the basic building blocks of carbohydrates – to link to form disaccharides, but also as the chains continue to elongate the complexity increases enormously, introducing differential modifications, side chains, branches or the addition of sulfate groups, for instance.
With such a powerful technology, Momenta ought to be able to build a nice business for itself through a series of carefully crafted, product-focused partnerships. But the company wants more: It also intends to develop its own drugs – new therapeutics as well as enhanced versions of existing products. Armed with the ability to sequence complex sugars, company scientists will also be able to analyse how specific sequences and changes in structure affect the properties of drugs.
Fully characterized heparin chain. Image courtesy Momenta Pharmaceuticals Inc.
For instance, they’ve used the technology to understand the relationship between structure and activity for heparin, a complex, sulfated polysaccharide that blocks the formation of blood clots. “Then we designed a product with improved clinical properties for specific indications,” Venkataraman said. (You can read the details in the following paper: Rational Design of Low-Molecular Weight Heparins With Improved In Vivo Activity. Sundaram et al.; Proceedings of the National Academy of Sciences, 100:651; January 21, 2003.)
In fact, the company is now developing an anti-coagulant drug candidate with superior efficacy, safety and pharmacokinetic properties.
Moreover, understanding sugars should open the door to new therapeutic approaches for treating cancer and other diseases. Because the sugar groups on cancerous cells are very different than those found on normal ones, for instance, it’s possible to analyze those differences and determine the effects that those changes have on disease biology. As well, since cell surface sugars modulate the access of growth factors to the cell, engineering sugars might be a way to block multiple pathways simultaneously in cancer cells, Venkataraman said.
It’s also possible to design small molecule drugs that mimic the actions of biologically important carbohydrates – or so the founders of Rockville, MD-based GlycoMimetics Inc. hope. The fledgling company was spun out of GlycoTech Corp. to develop the latter’s platform technology for designing and screening carbohydrate mimics – itself the result of a seven-year collaboration between GlycoTech and Ciba-Geigy Ltd. (now Novartis AG).
The young firm’s first goal is to develop compounds with anti-inflammatory activity, especially compounds that inhibit selectins (also known as cell adhesion molecules), which are a group of carbohydrate-binding proteins that regulate leukocyte trafficking to sites of inflammation. “We’ve developed screening methods and acquired technology rights for selectin inhibitors,” explained GlycoMimetics’ CEO Rachel King. These selectin inhibitors are glycomimetics developed by John Magnani, the firm’s scientific founder, under the Ciba-Geigy collaboration. “He determined the bioactive conformation of a carbohydrate when it enters into the binding site of the relevant protein,” King said, then designed mimics that exhibit higher binding affinity than their native counterparts. And what really sets these mimetics apart, she added, is the fact that they are hetero-bifunctional.
GlycoMimetics is also developing carbohydrate mimics for treating infectious diseases. “It’s well known that a number of infectious agents bind to cells through carbohydrate interactions,” King said. In cystic fibrosis, for example, “Pseudomonas aeruginosa binds to a particular carbohydrate in the lung epithelium.” Interestingly, this carbohydrate “has the same structure as one that GlycoTech was modeling for many years,” she added, making it a natural drug development candidate for the company.
While many companies (including, but certainly not limited to, those mentioned above) are focused on developing carbohydrate-based drugs or glycomimetics, others have turned their energies towards improving the glycosylation patterns on protein-based therapeutics.
As alluded to earlier, even highly purified recombinant proteins exist as a mixture of different glycoforms. For example, it’s been reported that MedImmune Inc.’s humanized monoclonal antibody Synagis consists of seven or eight glycoforms. Obviously, a certain amount of microheterogeneity is acceptable to the FDA, but what if the medicine consisted of purely homogeneous antibody molecules, each one identical to the others in terms of the particular carbohydrate groups attached to their amino acid backbones? Would this version be more potent or efficacious than the other?
That’s one of the questions that’s intrigued researchers at privately held GlycoFi Inc. – and one that their technology might be able to answer. The Lebanon, NH company was founded in 2000 by Dartmouth College professors Charles Hutchinson and Tillman Gerngross to improve the capacity and cost of producing therapeutic proteins, while simultaneously enhancing their efficacy and safety.
Realizing that the standard production system for glycosylated recombinant proteins – Chinese hamster ovary (CHO) cells – is expensive and inefficient, GlycoFi turned instead to yeast and other fungal expression systems, which are inexpensive to operate and can make proteins at very high concentrations. However, they first had to overcome one major obstacle: Yeast don’t glycosylate proteins in the same way that mammalian cells do. In fact, yeast introduce different inter-sugar linkages and add large amounts of mannose, which can be immunogenic in humans.
GlycoFi’s fungal cell lines, each of which is able to produce a desired glycoprotein in a specific glycoform.
Image courtesy GlycoFi Inc.
To accomplish their goal, the scientists reengineered the glycosylation pathway of the yeast Pichia pastoris (already an industrial workhorse) so that it correctly produced human glycoproteins. They achieved this remarkable feat by first eliminating the endogenous yeast glycosylation pathways, then introducing five active eukaryotic glycosylation enzymes. As a result, the yeast were able to produce a complex human N-linked glycan. Moreover, the humanized yeast yielded essentially homogeneous glycoforms. (You can find the details in the following paper: Production of Complex Human Glycoproteins in Yeast. Hamilton et al. Science 301:1244; August 29, 2003.)
“We still need to further humanize this pathway,” explained Gerngross, including introducing the capability for adding galactose and sialic acid residues, which are common (but not ubiquitous) on human proteins. “We’re working on that.” But the system as it stands is already appropriate for making antibodies, “which don’t have sialic acid residues and only partial or no galactose residues,” he added.
Using its combinatorial genetic library, which represents tens of thousands of fungal strains, GlycoFi has developed a wide variety of fungal cell lines, each able to produce a desired glycoprotein in a specific glycoform. By manipulating the particular sugars that are added to a protein, it should be possible for researchers to understand which moieties contribute to the protein’s efficacy – and how.
“Erythropoietin (EPO) isolated from humans has up to 100 different glycoforms,” Gerngross said. The protein has three glycosylation sites and all the permutations are in the mix. But some glycoforms are better than others. Now we can make EPO in one specific glycoform and compare it with another specific glycoform to determine which is the better therapeutic.”
While GlycoFi has chosen to address the problem of incorrect glycosylation patterns by humanizing yeast, Neose Technologies Inc. has taken a different route altogether. For, it’s not just yeast that present a challenge. No other production host for making recombinant proteins -- including insect cells, transgenic plants and transgenic animals -- adds sugars in a human-identical fashion. In fact, even CHO cells aren’t perfect, for they don’t necessarily complete the entire glycan (complex carbohydrate) structure in their rush to churn out product. Neose’s approach to fixing this problem is to modify the recombinant proteins after they are made by using its library of recombinant enzymes to remove incorrect sugars and then replace them with the naturally occurring versions. The technology, termed GlycoAdvance, is also used to add missing sugars to incompletely glycosylated proteins. (The Signals article, “Glycosylation Matters,” explores the differences among glycosylation patterns from these various host systems in more depth.)
The Horsham, PA company has parlayed its core technology into drug development, too – and it’s already identified its first candidate, a longer-acting glycoPEGylated EPO. This enzyme-based technology enables the selective addition of polyethylene glycol to sugar chains. Importantly, the PEG is linked to glycans that are distant from the protein’s active site, thus avoiding interference problems. According to executive VP David Zopf, “The principle of adding PEG is not new, but our way of adding it is. PEG is going onto existing glycan chains. There are a defined number of [glycan-binding] sites, and they are located in such a way that the glycan chains point away from the active site of the protein. We don’t want PEG attached in such a way that there is steric hindrance when the protein binds to its receptor.” He added that this technology yields “very uniform production of biologically active molecules.”
So how does Neose’s glycoPEGylated EPO differ from Amgen Inc.’s second-generation EPO, Aranesp? “Aranesp has five N-glycosylation sites compared to the three sites in the native (and recombinant) molecule,” Zopf explained. “Scientists changed the amino acid sequence to add these additional glycosylation sites. At each site, the expression system adds tetra-antennary chains, which adds eight negative charges. These changes increase the charge and the molecular weight of the molecule by a few thousand Daltons.” As a result, Aranesp has a longer serum half-life than Epogen. “It’s not completely clear why,” he continued, “but it’s partially due to decreased renal clearance and decreased receptor-mediated clearance.”
On the other hand, Neose’s glycoPEGylated EPO “has the original three N-glycosylation sites. We add a single PEG at each of these.” Thus, Neose’s molecule has less negative charge than Amgen’s, but a somewhat lower molecular weight. “The PEG we put on more than makes up for the carbohydrate bulk in Aranesp,” Zopf said.
Recent history has already demonstrated the success of PEGylated protein-based therapeutics – including PEG-Intron, Pegasys and Neulasta. Given that, plus the FDA’s comfort level with these sorts of improvements, the many biologics that will be coming off patent in the next 5-10 years “will be excellent candidates to be re-introduced into the market” as improved, PEGylated versions, he said.
There are other ways to harness the power of Neose's technology, too. The company is also using its technology to improved the function of antibodies and glycoproteins by attaching new bioactive or functional components to them. “This is the same approach as glycoPEGylation,” Zopf explained. “We couple the new components through the glycan chains at a defined number of sites. We can pair molecules together, such as two glycoproteins, by setting up a linker.” This approach would allow the linking of two glycoproteins with different functions – an antibody coupled to a toxin, for instance. “Older linker technologies depend on chemical reactions with the polypeptide chain at particular amino acids.” Unfortunately, some of those amino acids may reside at the protein’s active site, which would reduce or even eliminate its biological activity. Neose’s glycoconjugation technology gets around that problem.
Thus, the “new generation” glycobiology companies – exemplified by those discussed in this article – are developing powerful technologies to unlock the mysteries of glycoproteins. They’re devising the means to finally reveal the complete sequence of sugars in large, complex glycoproteins, which will lead to highly productive analyses of structure-activity relationships as well as the ability to create entirely unique therapeutics. They’re also uncovering the many functions that cell surface carbohydrates play in biology – and from that knowledge, a clearer understanding of how the functions go awry in the course of disease.
Some researchers are even discussing a complete analysis of the glycome – all the sugars that a cell makes. But this is a gargantuan undertaking, indeed, for the glycome of a single cell type is thousands of times more complex than its proteome – which is, in turn, more complex than its genome. Moreover, it’s known that a given cell’s glycome is constantly changing in response to environmental cues. And, when you consider how many different cell types there are, the challenge of glycomics appears overwhelming.
Even the Consortium for Functional Glycomics, which is being funded by the NIH’s National Institute of General Medical Sciences, isn’t prepared to tackle the glycome. Instead, it’s concentrating on understanding the role of carbohydrate-protein interactions in cell-cell communication. It’s a functional approach (as the name states) – not a wide-spread effort to catalog every single sugar in a cell’s repertoire.
One day, glycomics may indeed come into its own – just as genomics did. But for now, glycobiologists have plenty of projects to keep them busy.
|